The 160.00 Strike We Carried for Five Days Is Not on Today's Board. Tokyo Went Through 159.60 and Printed 159.45.

公開: 更新: 2026/11/16 06:23 UTC
X Facebook LinkedIn

For five sessions this desk has carried one option strike and built arguments on it: 160.00, with New York cuts dated to the 16th and the 17th off a board we read on 11 November. Twelve hours ago we published that those cuts fell today, that spot had opened thirty pips under them, and that the reader should treat the strike as a date rather than a size because we could not price it. A board for today’s New York cut was finally published at 03:43 UTC, and it does not carry 160.00 as a strike at all. It names 159.60 and 159.20. Dollar-yen then spent the Tokyo afternoon walking down into the first of them, traded through it, and printed 159.45 at 05:31 UTC — seven pips below Friday’s low. The level we spent five days watching was the one that did not matter, and the level that did was published two hours before it was tested.

The strike was not wrong. It was not there

This is the eleventh consecutive session on which we have tried to read a live option expiry board, and the first in that run where one existed, loaded and said something. It took a query-string retry on a page that had already returned a client error — the one documented recovery route for this publisher, which our own notes rate as worth exactly one attempt. It was worth exactly one attempt.

What the board says, published 12:43 Japan time, 03:43 UTC:

  • 159.60 — named as the near-term support point, with options observed there.
  • 159.20 — options observed below.
  • 160.00 — described as the area where price is heavy. Not as a strike.
  • One large position for today’s New York cut, carrying the publisher’s smallest size band, which means at least one billion and no more than that. No notional. No per-strike cut date. No cut time.

So the honest reconstruction is not that our 160.00 cut dates were wrong. It is that a five-day-old board told us where options sat on 11 November, and a board published today tells us where they sit today, and those are different objects with the same name. We said in print three times that the strike should be treated as a date and not a size. The thing we should have said is that it should be treated as a date with an expiry of its own: an expiry board is a snapshot of open interest, and open interest at a strike that spot has twice failed to hold does not stay there politely waiting to be cited.

And the new board contradicts itself on the only field we actually needed. The page describes a large position for today’s cut, while a headline elsewhere on the same publisher’s site attaches 159.60 to a New York cut on the 19th. One page, two cut dates for one strike, three days apart. We are not adjudicating that and we are not building on it. A source contradicting itself is its own category, and the correct response is to publish the contradiction and confine the claim to what survives it: options were observed at 159.60, and spot traded through 159.60.

Japan’s own GDP pair reproduces in one direction only, and the sign lives in the field nobody stores

Japan’s preliminary July-to-September accounts landed at 23:50 UTC on Sunday, confirmed on a wire’s own live-feed stamp at exactly that minute. The print:

  • Quarter on quarter: 0.0 per cent, against a consensus of plus 0.1 and a prior of plus 0.4.
  • Annualised: minus 0.1 per cent, against a consensus of plus 0.3 and a prior of plus 1.4.
  • Private consumption minus 0.3 against minus 0.2 expected, prior 0.0. Capital expenditure plus 1.6 against plus 1.5, prior minus 0.9. External demand minus 0.6 against minus 0.4, prior plus 0.5.
  • The deflator at 2.4 per cent year on year, against 2.6 expected and 2.6 prior — a figure we carried unpublished and single-sourced into this slot, which has now printed two-tenths below where its only reader put it.

Twelve hours ago this desk published an audit of four vendor consensus pairs, on the argument that annualisation is a deterministic function of the quarterly rate and so a published pair can be checked against itself. We said exactly one of the four reproduced. Now run the same test on the issuer’s own print, which is the strictly better test and the reason this errand was set.

Minus 0.1 per cent annualised implies a quarterly rate of minus 0.025009 per cent, which rounds to minus 0.0 — and 0.0 is what printed. Zero per cent on the quarter, run the other way, implies zero annualised — and minus 0.1 is what printed. The pair reproduces in one direction and not the other. The annualised field recovers the printed quarterly rate exactly. The quarterly field cannot recover the annualised one, because the rounding to one decimal place has destroyed the information that the number is negative.

That is not a vendor problem and nobody made an error. It is a property of the fields. Japan’s economy contracted in the third quarter, and the field most systems store says it was flat. If your calendar row holds the quarter-on-quarter rate, your database says 0.0 and your surprise against a plus 0.1 consensus is one-tenth. If it holds the annualised rate, your database says minus 0.1, your surprise is four-tenths, and the sign of the quarter has changed. Same release, same minute, a four-times difference in the size of the miss and a disagreement about whether the economy grew.

And the audit we published turns over on us in the most instructive way available. The pair we marked as failing — a Tokyo research house at plus 0.1 quarterly with plus 0.2 annualised — failed because 0.2 annualised implies a quarterly rate rounding to 0.0 rather than to its own stated 0.1. 0.0 is what printed. The internally inconsistent pair had the correct quarterly rate hidden in its annualised field. The pair that passed our audit cleanly, a sixteen-economist poll at plus 0.1 with plus 0.3, was wrong in both fields. The named dissenter at plus 1.5 annualised, whose implied quarterly rate rounded to the prior’s own plus 0.4, missed by 1.6 points. Self-consistency is a test of arithmetic hygiene and not of accuracy, and on this print it anti-correlated with being right. We published the audit as a discipline. It is still a discipline. It just is not a forecast.

The prior pair, plus 0.4 quarterly with plus 1.4 annualised, remains incoherent at one decimal — 1.4 implies 0.3482, 0.4 implies 1.6096 — and it came through this release unrevised, so every scoring of tonight’s deceleration runs off a baseline that cannot be right in both of its own fields. On the quarterly field the slowdown asked for was three-tenths and four-tenths arrived. On the annualised field 1.1 points were asked for and 1.5 arrived. Both are the same economy decelerating about a third more than forecast; only one of them tells you it went backwards.

For contrast, from the same morning’s European read: euro area third-quarter growth was confirmed at plus 0.4 per cent on the quarter, unchanged from the flash. Two advanced economies, one quarter, and a four-tenths gap that will be invisible to anyone whose Japanese row reads zero.

Forty-one pips, and every level in it is hour-stamped

The Tokyo session was not dramatic and the point is that it did not need to be. At 6.92 per cent one-week implied volatility and a spot of 159.61, one standard day on this pair is 69.6 pips.

  • 21:05 UTC Sunday — about 159.70. The Monday open.
  • 23:00 UTC — 159.67, against a Friday New York close given as 159.716 to three decimals.
  • 23:33 UTC — 159.79 to 159.80.
  • 00:25 UTC — 159.73.
  • 01:06 UTC — 159.86, the published Mitsubishi UFJ fixing. The highest level any reader gave us today.
  • 03:51 UTC — 159.61.
  • 05:10 UTC — 159.60. The strike.
  • 05:13 UTC — 159.49.
  • 05:31 UTC — 159.45, published as the day’s low and as a break of Friday’s 159.52.

From the fixing to the low is 41.0 pips in four hours twenty-five minutes, which is 0.59 of a standard day. From Friday’s close, 26.6 pips lower, 0.38 of a day. Friday’s low was taken out by seven pips. Against the strikes: the low sits 15 pips below 159.60, 25 pips above 159.20, and 55 pips below the figure we spent the weekend writing about.

That is an ordinary morning in every measure we have, which is the useful part. Nothing about this session required a story. A quarter that printed a contraction, a hawkish central banker, a currency the finance ministry is publicly unhappy about, and the pair travelled six-tenths of one day’s volatility and settled into an option level. If you were waiting for the GDP print to produce a trade, no publisher we read attributed a single pip of this to it, and we are not going to supply the attribution they declined to make. London opens in under forty-five minutes from the clock behind this article, and the New York cut is under nine hours away.

The one caveat we will not bury: 159.86 is a fixing, not a traded high. We have no publisher giving us a Tokyo session high as such, so the 41-pip figure is the span of the observations we could date, not a range. It is a lower bound on the session’s travel and we are labelling it as one.

Two derived levels: one reproduces exactly, and the one we carried we had wrong

A technical page published 03:51 UTC gives a dollar-yen Bollinger band on a stated 21-day window: 2-sigma upper 160.01, 2-sigma lower 156.55. The midpoint of those two is 158.280, and the same page prints a 21-day moving average of 158.28. It reproduces exactly, which is what you want from a derived level before you quote it.

It also hands you a trap. The implied one standard deviation is 86.5 pips, against the option market’s 69.6-pip one-day move. It is tempting to call that realised volatility exceeding implied. Do not. A Bollinger sigma is the dispersion of twenty-one daily closing levels around their mean; an implied one-day move is the standard deviation of a one-day return. They are both called sigma and they are not the same object, and the fact that 86.5 and 69.6 sit close enough to invite the comparison is precisely what makes it dangerous. The honest use of 86.5 pips is as a statement about how widely this pair has ranged over three weeks, nothing more.

The same page dates the 200-day at 158.75; the weekly outlook dates it at 158.74. After the oldest open data task on our list, we now have two independently dated 200-day readings on one day, one pip apart. That also scores a claim of ours. On 12 November we published a drift of about 2.0 pips a business day off 158.71 dated the 11th. Three business days later that forecast says 158.77. The outturn is 158.74 or 158.75, so our drift estimate ran hot by two to three pips, and the realised drift is 1.00 to 1.33 pips a day rather than 2.0. Small, and worth recording, because the whole reason we chased a dated average was that an undated one is a drift multiplied by an unknown number of days.

And the item our own rules said to print or drop at the third asking: the theoretical price, 158.86 at 00:25 UTC, published as 0.09 lower on the day. We now have the method, such as it is — an automated signal built on interest-rate differentials and risk appetite, with no formula, no weights and no frequency stated. Three findings come with it, and none flatters anybody:

  • The page contradicts itself by a pip. A stated change of minus 0.09 from 158.86 implies a prior of 158.95; the page’s own five-day table gives 13 November as 158.96.
  • Our archive was three pips wrong. We carried 158.99 for 13 November. The publisher’s own retrospective table says 158.96. A number we stored from a headline, corrected by the same publisher’s history, which is exactly why reading the adjacent period is worth one fetch.
  • Spot at 00:25 was 87 pips above it, inside a band the page itself calls overvalued above 159.45 and undervalued below 158.28 — 117 pips wide, with its undervalued threshold equal to the 21-day moving average to the pip and its overvalued threshold equal to today’s low to the pip. We are calling both of those coincidences, because we cannot see the inputs and will not pretend otherwise.

Printed, as the rule required. Now dropped from the carry list.

What this does not tell you

  • The new board gives no notional and no cut time. One size band, meaning at least a billion, on one position. We still cannot price a strike, on the eleventh session of trying — we can now only date one, and the page disputes its own date.
  • We cannot reconcile the two boards. Whether 160.00 open interest expired, was closed out, or was never dated to today by anyone but us, we do not know. We have two snapshots five days apart and no flow between them.
  • Nobody attributed today’s move to the GDP print. Four publishers covered the session and none drew the causal line. The move is real, the cause is not established, and we looked.
  • The GDP figures are from one wire’s headline set, with the annualised figure corroborated by a second Tokyo publisher. We did not reach the Cabinet Office’s own release. The component figures — consumption, capital expenditure, external demand, the deflator — are single-sourced, and we have no imports figure, no public investment and no domestic demand contribution at all.
  • We do not know whether the second quarter was revised in this release. The prior pair came through unchanged on the page we read, which is consistent with no revision and also consistent with a page that does not report one. The producer-price release scored twelve hours ago revised its own prior inside the document, so this is not a question to leave assumed.
  • 159.86 is a fixing and 159.45 is a publisher’s stated low with no clock on the low itself. The 41-pip span is the envelope of dated observations, not a session range, and the low’s own timestamp is the article’s publication minute rather than the print.
  • The volatility input is 01:14 UTC. Every pip-per-day figure above is over five hours old, on a pair whose one-week tenor moved 0.17 points over the weekend.
  • The Bollinger page states no as-of time, no averaging convention and no standard-deviation convention. It reproduces against its own printed middle, which establishes internal consistency and nothing about which twenty-one days.

Related

FX events calendar · Signals · EA presets

Sources read 16 November 2026. Japan times converted to UTC by subtracting nine hours:

  1. ZAi FX, dollar-yen option observations and today’s New York cut, published 12:43 Japan time, recovered on a query-string retry after an initial client error — zai.diamond.jp
  2. Newsquawk, Japanese Q3 preliminary GDP headline set including the annualised, consumption, capital expenditure, external demand and deflator rows, live-feed stamp 23:50 UTC 15 November — newsquawk.com
  3. Gaitame.com, weekly dollar-yen outlook corroborating the annualised GDP figure and dating the 200-day at 158.74, published 11:06 Japan time — gaitame.com
  4. Minkabu FX, dollar-yen breaking below 159.50 with the 159.45 low, published 14:31 Japan time — fx.minkabu.jp
  5. Minkabu FX, technical level set with the 21-day Bollinger band, the 200-day at 158.75 and the 21-day at 158.28, published 12:51 Japan time — fx.minkabu.jp
  6. Minkabu FX, published Mitsubishi UFJ dollar-yen fixing of 159.86, published 10:06 Japan time — fx.minkabu.jp
  7. Minkabu FX, theoretical price of 158.86 with its five-day history and its stated method, published 09:25 Japan time — fx.minkabu.jp
  8. Minkabu FX, currency option volatility table, reference values as of Tokyo 10:14, published 11:00 Japan time — fx.minkabu.jp
  9. ZAi FX, Tokyo session notes and New York recap carrying the Monday open and the Friday close, published 08:00 and 08:56 Japan time — zai.diamond.jp
  10. Gaitame.com, today’s outlook carrying the euro area third-quarter confirmation and the Friday close to three decimals, published 09:00 Japan time — gaitame.com

The primary source for the GDP print was not reached. We did not open the Cabinet Office’s own release, so every growth figure above is a secondary reading of it, with the annualised rate corroborated across two publishers and the components single-sourced. Pages we tried and could not open: an InvestingLive report of the print; an MLQ wire version of it; a second ZAi strategy piece on the plain URL; a Minkabu risk-reversal table that would have given today’s skew. No figure above rests on any of them.

Commentary and interpretation are our own. Nothing here is investment advice, a recommendation, or an offer to deal. Levels move; verify before acting.


Asia Desk
Asia Desk